Prospective development of a patient-reported outcomes instrument for desmoid tumors or aggressive fibromatosis.

Thirty-one patients (mean age, 44 years; standard deviation, 13 years; 77% female) completed the CE phase, and the demographics are described in Table 1 . The majority of the patients had at least a college degree (71%) and were symptomatic (84%). Tumor site and type varied among all patients. Interviews took approximately 60 minutes to complete. The 31 interviews were split into 3 “groups” to assess saturation. Group assignment was determined by the chronological order of interview completion. Saturation details for patients and the total number of new concepts that appeared in each interview group are detailed in Table 2 . The new appearance of a concept was identified by an X in the transcript group column where it first appeared. Most of the coded concepts were identified in the first group. Because no new concepts were discovered in the second group and only 1 was discovered in the third group (ie, renal/kidney failure), it was determined that saturation was achieved. A total of 33 unique symptoms were identified ( Table 2 ). Those most frequently mentioned included “disfigurement/altered appearance,” “nerve pain,” “decreased range of motion,” “fatigue,” and “nausea.” Twenty unique impacts were elicited from patients, with “fear,” “difficulty sleeping,” “concern about lack of knowledge among health professionals,” “anxiety,” and “ongoing medical uncertainty” cited most frequently. Four key symptom domains (ie, Location [1 item], Pain [7 items], Physical Function [3 items], and Vitality [5 items]) and 5 impact domains (ie, Appearance [2 items], General Impact [1 item], Physical Function/Mobility [6 items], Psychological [6 items], and Sleep [4 items]) were identified, and they resulted in a 35-item draft instrument. A 24-hour recall period was selected for the symptom domain, whereas a 7-day recall period was selected for impacts. The 0 to 10 severity NRS (ie, from “none” [0] to “as bad as you can imagine” [10]) was selected for all symptom items with the exception of the location item, which asked patients to indicate the location(s) of their DTs from a list. A 0 to 10 NRS was also used for impact items; however, the scale anchors varied for amount (ie, “how much”; from “none” [0] to “a great deal” [10]), frequency (ie, “how often”; from “never” [0] to “all the time” [10]), satisfaction (ie, from “not at all” [0] to “as much as you can imagine” [10]), and severity (ie, from “never” [0] to “as bad as you can imagine” [10]). CIs were conducted with 15 patients (mean age, 45 years; standard deviation, 12 years; 73% female) independent of the CE sample across 3 phases. Only 1 patient had an education level of high school or less, and most were symptomatic (93%). As with CE, tumor site and type varied. Table 2 contains demographic and clinical information for the CI participants. Interviews took approximately 60 minutes to complete. After phase 1 of the CIs (n = 5), an item on “swelling in other areas” was added per patient suggestion. Instructions for the impact scale were modified and separated into frequency and disturbance sections to enhance clarity. In addition, a 5-point verbal descriptor scale (ie, “none of the time,” “a little of the time,” “some of the time,” “most of the time,” and “all the time”) was included for the impact items that included frequency (ie, how often) to be evaluated against the 0 to 10 NRS in phase 2. Upon the conclusion of phase 2 of the CIs (n = 5), a decision was made on the basis of patient feedback to remove mention of attributing symptoms to DTs in the instructions. Six items were removed because of irrelevancy to DTs or redundancy with similar questions (ie, “zapping pain,” “muscle ache,” “throbbing pain,” “worn out,” “impact of difficulty sleeping,” and “difficulty bending, lifting, or stooping”). The “swelling in other areas” item that was added after phase 1 was deleted on the basis of patient feedback. In addition, examples in the “moderate activities” item were revised (ie, playing with children and taking a long walk) per patient suggestion. Patients preferred the 5-point verbal descriptor scale for frequency-based impact items. As such, the 0 to 10 NRS was removed for these questions. Per patient suggestion, an item to assess “weakness around your tumor” was added to compare with the “muscle weakness around your tumor” item in phase 3. Phase 3 of the CIs was completed among 5 additional patients. The “weakness around your tumor” item added after phase 2 was removed because patients preferred the specificity of “muscle weakness.” In addition, the “worst feeling of tiredness” item was removed because of redundancy with preferred items. The resulting questionnaire includes 28 items that capture symptoms and impacts related to DTs. Table 3 includes a summary tracking matrix of all items that were modified and/or removed, including the justification for any modifications and the phase during which the changes were made. The final version is termed the Gounder/DTRF Desmoid Symptom/Impact Scale (GODDESS) and includes symptom (11-item) and impact (17-item) scales (available to all with a material transfer agreement). Examples of GODDESS items are included in Figure 1 .

We sought to include adult patients who had (either at the time of the study or previously) localized or multifocal DTs affecting a variety of anatomic locations. Patients were eligible if they were aged 18 to 75 years and could speak English. The study population also included patients with familial adenomatous polyposis, a genetic condition that is correlated with a higher occurrence of DTs. Patients were ineligible if they were currently enrolled in a therapeutic clinical trial, were physically unable to participate in a 60-minute phone interview, or were affiliated or had a family member affiliated with one of the following: the US FDA or any other government agency that approves medications, an advertising agency, a marketing research company, or a pharmaceutical or biotechnology company. The New England Independent Review Board reviewed and approved the study protocol. Written informed consent was obtained from patients before online screening. Five independent, senior academic physicians with expertise in desmoid surgery or medical oncology were interviewed to better understand the disease process and their perspectives on symptoms, signs, and impacts on patient lives. Two moderators trained in qualitative methodology and content validation interviewing conducted the interview sessions, with a single moderator per patient. To ensure consistency between and across interviews, the content and process of each interview were shared with all team members. Cisco WebEx software was used to conduct and audio-record all interviews. Patients were free to discontinue their participation at any time. CE interviews began with the moderator asking patients to spontaneously identify symptoms and/or impacts that they attributed to their DTs. Patients were then presented with a list of symptoms and impacts that was developed through a review of the literature and expert consultation. The patients were asked if they recognized items from the list that they did not mention during the initial part of the interview. For the final portion of the interview, patients were asked to rate the level of disturbance for each of the identified symptoms or impacts on a 0 to 10 numeric rating scale (NRS), where 0 indicated “not disturbing at all” and 10 indicated “extremely disturbing.” Supporting Table 1 contains a list of sample CE interview questions. De-identified transcripts were made from all CE interviews. The primary goal of transcript analysis was to organize and catalog patient summaries of symptoms and impacts reported during the interviews via content analysis and ATLAS.ti software. 21 A custom code book was created on the basis of participants’ demographics, information about their DT diagnosis and treatment, personal descriptions of desmoids, and the symptom and impact data capture forms. Each transcript was analyzed to determine whether the symptom or impact was mentioned by each respondent on the basis of content analysis of their verbatim responses and whether this mention was spontaneous or required recognition from the list. As part of this process, patient statements reflecting similar concepts were grouped together under the same code. For example, patient statements of “I have trouble sleeping through the night” and “sometimes I can’t fall asleep” would both be included within the theme “difficulty sleeping.” The analyst’s discretion was used to make these decisions. In cases in which this categorization was unclear, moderator discretion or team consensus was used. The frequency of symptoms and impacts was calculated as well as ratings of the average disturbance of symptoms and impacts. Saturation of concept was defined as the point at which additional patient interviews did not contribute unique concepts or information. 18 CE interview results were used to develop a draft PRO questionnaire. The response options, recall period, and PRO measure formatting were selected in accordance with FDA recommendations. 17 Before the start of each CI, the moderator asked each patient to log into a WebEx conference line and complete the questionnaire electronically. Upon measure completion, the moderator conducted the interview. The CI guide was developed in accordance with best practice guidelines for conducting CIs when PRO instruments are being developed for use in clinical trials, 19 and it included interview probes that addressed patient comprehension of (1) instructions, (2) instrument items (ie, the meaning of the specific symptom, functional impact, and/or other aspect of health status), and (3) response options, including how patients selected their response and interpreted different response units on the scale (eg, what a 0 meant on a 0–10 NRS). Patients were also asked to identify any areas that they found to be confusing, problematic, or irrelevant to their experience. Audio recordings of the CIs were used by the project team to create detailed notes from each phase of CIs that included information from every patient about the meaning of each section/item in the assessment. After each phase of CIs, any problems or concerns noted by the interviewers were summarized for review by the project team. This review allowed for a quick assessment of patient comprehension and revision of the instrument to facilitate subsequent CI phases.

To date, disease-specific PROs approved for use in the regulatory setting have been established only for prostate cancer, 22 non–small cell lung cancer, 23 tenosynovial giant cell tumors, 24 and myelodysplastic syndrome. 22 Here, we describe the feasibility and prospective establishment of content validity of the first PRO instrument for DTs. GODDESS is a 28-item questionnaire that meets FDA regulatory requirements for a disease-specific (content validation) PRO instrument and is currently undergoing prospective psychometric validation in 2 ongoing, prospective, pivotal registration trials in DTs ( NCT03785964 and NCT03459469 ). After the establishment of psychometric properties consistent with FDA guidance, GODDESS may become a new regulatory endpoint in clinical trials and an important tool in the routine management of patients with DTs in the clinic. Before our work, clinicians often described pain and functional loss as the symptoms that affected patients with DTs. Our study is one of the first to provide a detailed window into the myriad of symptoms and psychosocial impacts of this disease on the lives of patients. 5 As expected, the symptoms and impacts preliminarily show variation by anatomical location, as seen with abdominal desmoids, which cause nausea and early satiety. In addition to pain and functional impairments, health care workers should also address fatigue, insomnia, anxiety, fear, and body dysmorphia. Psychosocial impacts that may be unique to DTs are the frustrations of having a locally infiltrative tumor that is neither malignant (rarely fatal) nor benign and the difficulty in communicating this to family and society at large. Lastly, this study demonstrates the feasibility and challenges of developing a disease-specific PRO for a rare disease. Industry, regulatory agencies, patient advocacy, and academia recognize the importance of PROs in drug development; however, there are many barriers to successful development and implementation. The first barrier is the fact that the prospective development of a disease-specific instrument following FDA guidance is labor-intensive, time-consuming, and costly. 17 , 23 , 25 Although this is feasible for common cancers, such initiatives are extremely challenging for rare cancers (or diseases), which now constitute 25% of all malignancies. For rare cancers or diseases, the challenges include 1) the identification of stakeholders (ie, academia, industry, and patient advocacy) who will lead the development effort, 2) the timely engagement of regulatory agencies such as the Clinical Outcome Assessment Qualification (Office of Hematology and Oncology Products) program within the FDA’s Center for Drug Evaluation and Research to obtain guidance, 3) the acquisition of research funding support, 4) the identification and accrual of patients with rare cancers requiring collaboration across multiple institutions, and, lastly, 5) subsequent validation of the PRO tool in prospective studies. 26 Our efforts required successful collaboration among academia, patient advocacy, and industry. The research and development of this instrument were funded by the Desmoid Tumor Research Foundation, a nonprofit patient advocacy group. The patient advocacy group highlighted this study at patient meetings and through its online portals. Lastly, collaboration with academia to involve physicians with DT expertise was essential. Although HRQOL instruments such as the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 14 and the MD Anderson Symptom Inventory 16 are routinely implemented in oncology trials, these are not used as primary endpoints in the regulatory setting. Although European regulators (ie, the European Medicines Agency) are more likely to expect and include these instruments as secondary endpoints in label claims, these instruments currently do not satisfy FDA requirements for qualification for use for a label claim because they are not disease-specific instruments. 17 , 25 This is illustrated in the label claims for new oncology drugs approved between 2006 and 2013: 14 of 42 new drugs had PRO-based claims in Europe, and only 1 of 43 did in the United States. 27 In conclusion, we describe the methodology, feasibility, and challenges of developing a disease-specific PRO for rare tumors. GODDESS is now translated into Spanish, Dutch, French, Italian, German, and Japanese and is available as an exploratory endpoint for further research.

Desmoid tumors (DTs) are locally aggressive connective-tissue sarcomas that have high morbidity and low mortality. 1 These are rare or orphan cancers with an annual incidence of 1000 patients in the United States. The disease predominantly affects young adults and can arise in any anatomic location but favors the extremities, joints, and abdomen. Depending on its location, patients can present with pain, a loss of range of movement or immobility, bowel obstructions and/or perforations, hydronephrosis, and a host of other symptoms. Treatment for desmoid tumors can include a wait-and-watch strategy, surgery, ablation, systemic therapies (cytotoxic, hormonal, and tyrosine kinase inhibitors), and/or radiation in appropriately selected patients. 2 Prospectively conducted clinical trials in DTs use standard endpoints such as Response Evaluation Criteria in Solid Tumors (version 1.1) response rates and progression-free survival to measure treatment efficacy. 3 These surrogate endpoints for overall survival may not be appropriate for a disease with low mortality and importantly fail to capture whether treatments truly improve symptoms and/or affect daily living. To date, there are limited data on the qualitative impact among patients affected by DTs. 4 , 5 Patient-reported outcomes (PROs) are direct measurements of patient experiences without any filtration or interpretation by a clinician or health care worker. 6 – 8 Measuring a patient’s symptoms and function is an additional dimension or endpoint that qualifies traditional endpoints such as response rates and/or overall survival. PRO measures enable patients, families, and clinicians to make rational, transparent, and patient-centered decisions weighing the impact of treatments on survival, quality of life, side effects, and financial burdens. The value of integrating PROs for symptom monitoring was recently demonstrated in a large, randomized trial of various chemotherapies, which showed significant improvements in quality of life and overall survival in comparison with the standard of care. 9 , 10 Similarly, integration of electronic PROs into the clinic for the routine management of patients is feasible and appears to improve the quality of care for patients and physician satisfaction. 11 Although there has been a significant charge to incorporate PROs into routine care and clinical practice, 7 , 9 , 10 the capture of this subjective information has been particularly challenging in oncology for a multitude of reasons. 12 We sought to prospectively develop a novel regulatory and clinical trial endpoint to characterize the subjective experience in patients with DTs, an ultrarare cancer. Although a recent review indicated the need for a health-related quality of life (HRQOL) tool for patients with DTs, 13 an important step in the regulatory context is to ensure that the PRO measure is uniquely developed for DTs to complement existing HRQOL instruments such as the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30, 14 the Patient-Reported Outcomes Measurement Information System, 15 and the MD Anderson Symptom Inventory, 16 which are not specific to a disease or condition. We sought to develop a PRO tool that captures desmoid-related symptoms and impacts in accordance with the 2009 guidance for industry from the US Food and Drug Administration (FDA): Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling . 17 Therefore, the objectives of the current study were 2-fold. The first objective was to prospectively explore and understand the symptoms of adult patients living with DTs, their experience with treatment, and the impact of the disease on their lives via concept elicitation (CE). 18 The second objective was to conduct cognitive interviews (CIs) 19 , 20 in a second cohort of DT patients to assess patients’ understanding of the instructions, items, and response scales of the instrument and make additional refinements toward establishing the content validity for a PRO for DTs that could be used as a clinical trial endpoint.